WAVELET BASED EEG SIGNAL SEIZURE DETECTION USING SVM CLASSIFIER
R.Shantha Selvakumari, V.Santhakumari, T.Menaka Abstract: Epilepsy is a brain disorder that causes people to have recurring seizures. About 50 million people in the world are diagnosed with epilepsy. Seizure detection is very important for the diagnosis and rehabilitation of the epileptic patients. This paper presents a wavelet-based seizure detection method with high sensitivity, accuracy and specificity. This method is applied to two different groups of EEG signals: 1) healthy (Normal) EEG signal; 2) Epileptic (Seizure) EEG signal. Wavelet decomposition is used to decompose the EEG signal into five bands. Effective features such as power, fluctuation index, coefficient of variation …show more content…
Introduction
Epilepsy is a common neurological disorder characterized by seizures. Epileptic seizure is a change in behavior due to sudden, brief excessive electrical discharges of neurons. Generally, Electroencephalogram (EEG) signal analysis is used for assessing brain disorders. In this paper DWT is used for the detection of seizure. EEG signal can be analyzed by using both time and frequency analysis. Time Frequency analysis [1], [2], such as discrete wavelet transform (DWT) is very effective for analyzing nonstationary EEG signals. Effective features should be selected for accurate detection of seizure. A number of wavelet features such as relative energy [3], entropy [4], mean [4], lacunarity [5] etc., are analyzed. A binary classifier is required for the classification of EEG normal and seizure signals. A plenty of statistical techniques are used for binary classification. Some of the most popular techniques include linear Discriminant Analysis (DA) and non-parametric statistical methods like Neural Networks. SVMs are promising statistical non-linear, non-parametric classification technique and it gives good results for medical diagnostics and is a powerful tool for pattern recognition